Skills

Execution modes

Fixed, recommended, auto, user choice and compare — and when each is right.

Availability note — these docs describe both working and planned product areas. Account authentication, administrator-managed AIHK/AICA tenant regions and keys, signed-in Composer runs and public Guard scans are connected. Skills, Memory, Agents, Workflows, Apps, billing, the public SDK/CLI and the unified API are previews unless a section explicitly says otherwise.


Roadmap and package-contract reference: this page describes the intended GenCMS runtime. The current deployment persists workspace records and drafts but does not execute marketplace Skills, index Memory for retrieval, publish packages, or run Agents, Workflows or Apps. See the deployed API reference for the routes available today.

Every Skill declares how a model is chosen. This is the field that makes a GenCMS Skill portable across providers instead of tied to one.

ModeBehaviourUse when
fixedRequires one named model.The Skill was tuned to one model's behaviour and breaks elsewhere.
recommendedGenCMS suggests one; the user may override.One model is clearly better but others are workable.
autoThe router chooses per run against cost and latency preference.The task is common and several models handle it well.
user_choiceThe user picks every time.Output style is a matter of taste, not correctness.
compareRuns on several models and returns all results.The comparison is the deliverable — audits, evaluations, visibility checks.

Testing across models before you publish

Publishers can run a Skill against every supported model and compare quality, schema validity, latency, credit consumption and error rate. Publishing a `recommended` mode without that evidence is guesswork.

The cross-model Skill test runner and gencms CLI are planned, not deployed. When this workflow ships, comparison results must come from actual evaluation runs; the documentation does not publish invented benchmark output.